**** BEGIN LOGGING AT Wed Mar 23 02:59:59 2016 Mar 23 03:40:33 Hmm, I bet we could remove the PKG_CONFIG vars / exports from bitbake.conf if we wrap pkgconfig-native, and create a pkgconfig-cross which emits a wrapper around pkgconfig-native that goes in crossscripts and sets the vars for target. Mar 23 03:40:41 might have to try that Mar 23 07:38:12 hi for a package there is dependency on another package which must be compiled after that, Mar 23 07:38:34 now I have added DEPENDS Mar 23 07:38:40 but still its not working Mar 23 09:59:00 is there some yocto helper or "canonical" command for recursively installing files from a directory in do_install()? imo it's usually a bad idea, but i need to work with some existing code. Mar 23 09:59:36 i see lots of recipes rolling their own with various 'tar -c ... | tar -x ...' variations Mar 23 11:14:57 Ulfalizer: So do you want to install actual source code to your target? Mar 23 11:16:58 Iirc there is a way to populate source packages somehow. Maybe you could use the archiver feature to build a source package of your recipe and install that to your target: http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#providing-the-source-code Mar 23 11:29:51 LetoThe2nd: Do you know, whether i can use ifupdown with systemd? I'm having a hard time finding a proper solution to change interface config during runtime (e.g. via ifconfig) in conjunction with systemd-networkd :/ Mar 23 12:08:19 Anticom: why would you want ifupdown and networkd? Mar 23 12:08:33 seems like saying you want bsd and linux kernels Mar 23 12:09:44 rburton: no i want ifupdown instead of networkd :) Mar 23 12:10:07 so… don't enable networkd? Mar 23 12:11:55 For the past hardware models we didn't use yocto and now there are two problems arising (If we want to switch to systemd): 1st our bussiness code is coupled to some tools like ifconfig, ifup, ifdown etc. 2nd: As i already said i'm having a hard time to find some runtime configuration equivalents to the way it works when using ifupdown. Hence i was was wondering, whether it would be possible to use ifupdown in conjuction with systemd (instead of networkd). I Mar 23 12:11:55 found this: http://patchwork.openembedded.org/patch/96013/ which made me wonder Mar 23 12:12:12 rburton: thought connman was the default then? Mar 23 12:12:43 Hello. Mar 23 12:12:45 depends on the image, some oe-core sample images use ifupdown, some use connman Mar 23 12:12:53 none use networkd and we disable it by default Mar 23 12:13:22 Random question, I have a u-boot recipe that should be using : SRC_URI = "git://git.denx.de/u-boot.git;branch=master;protocol=git" Mar 23 12:14:00 However, watching whats going on via trafshow it seems to be downloading the git archive from download.yoctoproject.org,http which is obv very slow Mar 23 12:14:05 am I missing something? Mar 23 12:17:53 it shouldn't be very slow, obviously Mar 23 12:18:03 its downloading from the mirror as it's generally faster Mar 23 12:18:16 halstead: ^ reports of our mirror being slow again :( Mar 23 12:19:01 Well, it may be that I'm nowhere near the mirror... Mar 23 12:19:14 Source Destination Protocol Size CPS Mar 23 12:19:14 ──────────────────────────────────────────────────────────────────────────────────────────────────────── Mar 23 12:19:14 download.yoctoproject.org,http 192.168.12.245,59175 tcp 28222K 30K Mar 23 12:19:46 Can I persuade it to use something else faster / more local to me? Mar 23 12:19:48 Hello everyone, I have a question regarding patching a package in do_patch task in *.bb file Mar 23 12:20:56 Is there a way to patch a pckage only for one template and not for all Mar 23 12:21:27 I tried using the TMPL variable that is passed to the build command but it is not visible in the bb file Mar 23 12:21:35 Any suggestions? Mar 23 12:21:36 pevster: unset MIRRORS and PREMIRRORS Mar 23 12:23:31 rburton: I don't actually set either explicitly in my local.con ; do you mean set both to null explicitly? Mar 23 12:25:35 pevster: take a look at poky/meta-yocto/conf/distro/poky.conf Mar 23 12:25:43 Those are probably the default mirrors you're currently using Mar 23 12:25:54 you could _prepend your own Mar 23 12:27:00 pevster: http://www.yoctoproject.org/docs/latest/mega-manual/mega-manual.html#var-PREMIRRORS Mar 23 12:31:31 rburton: Different topic: I'm having an issue with PACKAGE_FEED_URIS. https://gist.github.com/Anticom/31372445ad6b299ca108 Mar 23 12:34:45 rburton: oh nvm... bitbake -e just finnished: PACKAGE_FEED_URIS="/path/to/repohttp://my.ip:8000" ...didn't _append prepend a whitespace aswell? Mar 23 12:35:37 nah, thats += Mar 23 12:36:18 i love those kind of mistakes Mar 23 12:36:37 missing whitespace was the last thing i've expected to be the issue Mar 23 12:54:45 dragonator: what's a template in that context ? Mar 23 12:57:57 boucman_work: Template in this context is the image you want to build. Lets say I have configured two images to be created during a full build : A & B . And I want to patch the package only for image A Mar 23 13:00:28 dragonator: no, you can't... at least not simply. Mar 23 13:00:38 yocto build packages, then uses packages to build images Mar 23 13:00:58 but packages are built independantly of the image(s) they will be deployed on Mar 23 13:01:14 mhm...I understand Mar 23 13:01:19 maybe you could do two packages with different version of the same soft Mar 23 13:01:26 Ok. Then another question Mar 23 13:01:40 I think there is also a mechanism in image recipes to override files, but I never used it Mar 23 13:03:20 If my package depends on library, which is provided under different name, how can I tell it how to search for it ? Mar 23 13:06:14 ok, let me reformulate your qutestion, to make sure I understand... Mar 23 13:06:52 I have a binary on my target that needs a particular .so, that .so is provided by multiple packages, how do I deal with runtime-dependencies Mar 23 13:06:54 is that it ? Mar 23 13:16:37 yes :) Mar 23 13:17:14 I knwo about RDEPENDS variable , but it seems that It could not find the right .so Mar 23 13:17:39 RDEPENDS should depend on a package, not a library, so it should be the name of the package providing the library Mar 23 13:18:24 if multiple packages provide the library, you can have your package depend on a "virtual package" and the real packages providing that same "virtual package" but I never had to do that myself, so i'm not sure how it works Mar 23 13:21:08 Actually the problem isn't during do Mar 23 13:21:29 ops...During do_install task ,but during do_rootfs task Mar 23 13:21:47 My package bb file is processed without errors Mar 23 13:35:25 ok Mar 23 13:35:39 so what's the error ? Mar 23 13:35:51 (and what does your package put in RDEPENDS) Mar 23 13:39:24 my package depend on openssl (libssl.so.10 and libcrypto.so.10) Mar 23 13:40:38 ok, so you need to DEPEND on openssl (to make sure the packages are actually built) and that should be enough I guess... Mar 23 13:41:17 and the error is something like: Can install _ during do_rootfs Mar 23 13:42:06 Can copy the exact error in the moment ,because I started a clean build now :/ Mar 23 13:42:49 Can't* Mar 23 13:43:07 dragonator: can you pastebin the .bb of your recipe somewhere ? and what architecture are you targeting ? Mar 23 13:43:20 are you doing some multilib, mixing 32 and 64 bit ? Mar 23 13:48:26 yes, it is multilib Mar 23 13:48:37 but I haven't done it myself Mar 23 13:50:35 and I'm not sure If I can pastebin it somewhere because it's kind of work-related.Unfortunately I'm a bit stuck and seek external help :/ Mar 23 13:58:19 hmm, multilib works, but it's a can of worm... you seem to depend on the 32bit version of openssl, but maybe only the 64bit version is built ? I don't know... Mar 23 14:02:41 Well.Thank you for your time.I'll keep digging. Mar 23 14:04:03 dragonator: try running "bitbake lib32-openssl" and see if it helps Mar 23 14:04:16 (that's not a solution, but it will help diagnose the problem) Mar 23 14:04:32 I tried...It builds fine Mar 23 14:05:13 oh ? hmm Mar 23 14:05:18 :q Mar 23 14:05:22 ups.. Mar 23 14:39:46 hey, quick question Mar 23 14:40:06 when building an image recipe Mar 23 14:40:22 the final image is only built when all the QA issues are solved, correct? Mar 23 14:40:40 warnings can be ignored Mar 23 14:41:32 warnings are not fatal, errors are Mar 23 14:41:54 ok, but in case of QA error I noticed that the packaging is done regardless Mar 23 14:42:05 but the final image isn't built Mar 23 14:42:20 am I assuming the right thing? Mar 23 14:48:45 yes Mar 23 14:54:26 i suppose this is done to give you a chance to inspect what has gone wrong there :) Mar 23 14:55:41 jose1985: package qa happens in parallel to actual packaging Mar 23 15:11:44 Is there a good way to use a .deb file to install a component without having to write a behemoth of a recipe file? Mar 23 15:15:56 karobar: bin_package.bbclass ? Mar 23 15:15:57 http://git.openembedded.org/openembedded-core/tree/meta/classes/bin_package.bbclass Mar 23 15:16:51 thanks, will take a look Mar 23 16:16:19 Is there any convenient tool with minimal requirements to measure boot times? When i'm using systemd i can use "systemd-analyze plot" but now i'm looking for a tool that works both for sysvinit and systemd Mar 23 16:16:52 since i've ripped opengl out of my distro bootchart2 won't do it for me and i can't be bothered rebuilding almost the entire system to stuff it back in there Mar 23 16:19:26 i don't know any other tool than bootchart... but i'm suprised that bootchart needs opengl. It should only need it to render, not to record... Mar 23 16:19:37 you can't use bootchart2 without opengl? would't it only need opengl fo rthe gui bits which could be done separately from the acxtual measurement? Mar 23 16:19:42 my question exactly.. :) Mar 23 16:20:40 bootchartd doesn't need opengl at all, its a gtk2 app Mar 23 16:21:49 i want to override a kernel module (replace with out of tree module), but I get PR errors. I've tried adding the module via "PREFERRED_PROVIDER_virtual/kernel-module-" to my local.conf, but can't get it to work Mar 23 16:21:54 how can I accomplish this task? Mar 23 16:35:43 hm this is the message i'm getting: "Missing or unbuildable dependency chain was: ['gp-image-sdcard', 'bootchart2', 'python-pycairo', 'cairo', 'virtual/egl', 'libgl-mesa-dev']" Mar 23 16:36:01 rburton: thanks Mar 23 16:36:06 it makes sense actually Mar 23 16:36:28 And the only recipe providing libgl-mesa-dev i could find is http://git.yoctoproject.org/cgit.cgi/poky/plain/meta/recipes-graphics/mesa/mesa.inc Mar 23 16:37:02 And this .inc in turn has REQUIRED_DISTRO_FEATURES = "opengl" Mar 23 16:37:11 So what am i missing here if it's not opengl? Mar 23 16:39:42 my guess is that it's a bug in the bootchart recipe... Mar 23 16:40:14 when your distro has no opengl, bootchart can't be built because it doesn't separate the rendering part from the collecting part... Mar 23 16:41:20 boucman_work: kind of. i guess the issue is that bootchart2 recipe is providing pybootchartgui aswell Mar 23 16:42:00 yes, i Mar 23 16:45:22 kergoth: can i solve that issue with PACKAGE_EXCLUDE_append = " pybootchartgui" or something like that somehow? Mar 23 16:51:35 Dumb question. I the local.conf file I noticed that many additions leave a space after the first quotation mark such as DISTRO_FEATURES_remove = “ x11 wayland”. IS that space really necessary? Mar 23 16:52:16 riz__: for _append _prepend and _remove you most of the times want that Mar 23 16:52:24 += automatically inserts a whitespace Mar 23 16:52:31 the three others i just mentioned dont Mar 23 16:52:57 so if you do A="foo" \\ A_append="bar" you get A-->"foobar" which usually isn't what you want Mar 23 16:53:02 you usually want "foo bar" Mar 23 16:53:10 hence you need to do A_append=" bar" Mar 23 16:53:26 Ahhh. I see. Thanks! Mar 23 16:53:28 space is irrelevent for_remove Mar 23 16:53:31 _remove is word-based Mar 23 16:53:38 kergoth: fair enough :) Mar 23 16:53:45 its value will be stripped, split, and each word removed from the variable Mar 23 16:53:53 What about INHERIT? Mar 23 16:53:55 so the space doesn't hurt, but isn't needed in this context either Mar 23 16:54:04 INHERIT is a space separated list. what about it? Mar 23 16:54:05 btw kergoth will _remove only remove one occurence if there are multiple ones? Mar 23 16:54:20 Is it good practice to put a space with that? Mar 23 16:54:51 Anticom: no, it should be removing all Mar 23 16:54:57 riz__: that doesn't really make sense Mar 23 16:55:03 a space with what? Mar 23 16:55:04 OK Mar 23 16:55:07 what exactly are you doing to INHERIT? Mar 23 16:55:14 INHERIT += “rm_work” Mar 23 16:55:21 as i already said, the space in an _remove will have no affect whatsoever, regardless of the variable Mar 23 16:55:23 riz__: kergoth just told you it's most of the times about lists and there the elements are seperated with a whitespace Mar 23 16:55:32 look at the example i gave you just a moment ago Mar 23 16:55:34 as i just said, INEHRIT is space separated. if you don't include the separator, then it's going to break Mar 23 16:55:39 you usually want "foo bar" instead of "foobar" Mar 23 16:55:40 += adds it for you Mar 23 16:55:44 indeed Mar 23 16:55:53 it has a different meaning Mar 23 16:56:11 Understood Mar 23 16:56:39 kergoth: any clue whether i'm on the right track with my guesses concerning bootchart2 ? Mar 23 16:57:55 PACKAGE_EXCLUDE controls exclusion of packages from images, to a limited extent. it won't prevent packaging, and it defnitely won't remove dependencies. best to examine the bootchart2 recipe and see whats what, i haven't seen it, so couldn't say Mar 23 16:57:59 afaik, anyway Mar 23 17:12:28 hi guys, fast question: I've moved my build-dir in another place and now I can't build due to this error: Error, TMPDIR has changed location. You need to either move it back to /home/***/***/***-alsafree/tmp or rebuild Mar 23 17:14:11 how can I rebuild? Mar 23 17:14:30 delete tmp Mar 23 17:16:42 thank you Mar 23 17:17:28 Hmm, I wonder if we should include the system hostname in the sanity file in conf/ Mar 23 17:18:05 i was thinking chroot and container builds that might use a build dir, easy to forget to remove it manually Mar 23 17:18:21 not sure many folks care about that use case, though, so possibly not Mar 23 17:19:05 also chroots wont change the hostname unless you jump through other hoops / change it otherwise. containers at least can change it via that namespace Mar 23 17:31:27 hi guys Mar 23 17:31:47 I have an image that have another image inside Mar 23 17:33:08 to do that, I made the same thing the kernel.bbclass does, creating a function that copy the image from deploy dir and add the task dependency to the do_build from the other image Mar 23 17:33:38 but this image that is inside installs a package afw-initscripts that rprovides initscripts Mar 23 17:34:20 and the image container, would install the normal initscripts, cause it is my preferred_provider Mar 23 17:34:37 but it is not happening, both images are installing afw-initscripts Mar 23 17:35:06 any tips about what I'm doing wrong? Mar 23 18:27:09 Hi, can anybody help me using wic with the jethro release? I currently see a pythonexception when running do_rootfs of the image recipe: "ImportError: No module named pysh" Mar 23 18:27:17 if two layers provide the same bbclass, is there some way to prioritize one over another? Mar 23 18:27:41 (and yeah, that's horrible) Mar 23 18:28:17 Ulfalizer: yeah, through the layering system. Look at the BBLAYERS variable Mar 23 18:28:47 I imagine that the BBFILE_PRIORITY can solve this Mar 23 18:30:08 psidhu: BBLAYERS just seems to be a list from looking at the reference manual. nothing says classes from earlier ones take precedence. Mar 23 18:30:31 Ulfalizer: hm, I've always used it as the first to appear has higher precedence Mar 23 18:30:49 but maybe igor3 has a better solution for you Mar 23 18:31:00 don't think BBFILE_PRIORITY would help either Mar 23 18:31:16 it probably won't ever even look at one of the bbclasses Mar 23 18:31:55 psidhu: that's what PREFERRED_PROVIDER and BBFILE_PRIORITY are for i think Mar 23 18:32:16 ooh, I didn't notice that you're talking about classes, sorry Mar 23 18:37:20 Ulfalizer: From the manual: Through the use of the BBPATH variable, BitBake locates class files (.bbclass), configuration files, and files that are included with include and require statements. For these cases, BitBake uses the first file that matches the name found in BBPATH. Mar 23 18:39:30 psidhu: would still need to assume that the layers are processed in a certain order for that to be helpful Mar 23 18:39:56 feels pretty brittle :/ Mar 23 18:40:12 since BBPATH is built by the layers' layer.conf files (iirc) Mar 23 18:40:32 Ulfalizer: yeah, but it's a start, no? Adding to BBLAYERS should processes that meta layer first, and each meta layer's layer.conf should append itself Mar 23 18:40:38 thus creating order? Mar 23 18:41:01 that assumes layers are processed by going linearly through BBLAYERS Mar 23 18:41:08 likely, but hrrr... Mar 23 18:41:15 Ulfalizer: I think they are Mar 23 18:41:39 the clean thing to do would be to get rid of the other bbclass file, so i'll go with that Mar 23 18:41:43 or try to anyway Mar 23 18:41:47 haha okay Mar 23 18:41:58 thanks for the help anyway :) Mar 23 18:47:02 Ulfalizer: BBLAYERS controls layer.conf parse order, and since layer.conf does BBPATH += in nearly all cases, the BBPATH Order is controlled by BBLAYERS order, and class/config file parsing uses the first from BBPATH. Mar 23 18:47:21 Ulfalizer: mentor's setup scripts reorder BBLAYERS by BBFILE_PRIORITY for consistency, rather than relying on the user ordering it sanely Mar 23 18:47:22 heh Mar 23 18:47:26 keep meaning to submit that upstream Mar 23 18:48:11 kergoth: yeah, guessed that was the only way to control it. thanks. Mar 23 18:49:15 really hae to address it through documentation on your layer today, try to get the user to order it properly when they add it to bblayers.. course most users don' read docs, so you're basically hosed ;) Mar 23 18:49:52 they happen to be in the wrong order in this case too Mar 23 18:50:12 but i'll try to get the clean solution with a single bbclass file before i rearrange it Mar 23 18:51:05 often you can get away without overriding the class entirely, i.e. add a new class which is opted in, which uses anonymous python or whatnot to counteract what the other class did Mar 23 18:51:14 i've done that with the mel distro before to avoid copying classes from oe-core Mar 23 18:51:16 heh Mar 23 18:52:03 * Ulfalizer wants less entropy :S Mar 23 18:54:37 does anyone know how to override a kernel module and not run into packaging issues? Mar 23 18:55:33 what exactly do you mean by 'override a kernel module'? Mar 23 18:56:28 so say there's an in-tree kernel module called 'kernel-module-hello', and I have another recipe that also creates that kernel module and I want to overwrite the first one Mar 23 18:57:06 disable the in-tree module :) Mar 23 18:57:27 kergoth: I realize that's the easy solution, but I need to have it enabled for certain build dependencies in the kernel Mar 23 18:57:31 haha Mar 23 18:58:30 i doubt you're going to find a clean solution, most likely you'll have to hack something. you could use a different package name for yours, then use RPROVIDES/RREPLACES/RCONFLICTS to let the package manager replace the earlier installed package with yours as the later installed package, but bitbake may still get unhappy about both providing the same thing Mar 23 18:59:22 kergoth: hm, alright. thanks for the insight Mar 23 19:03:27 Has anybody used wic with the jethro release? As written above, I see some python exceptions when running wic on Ubuntu 16.04. Mar 23 19:51:23 RP: I would like us to enable world build for musl on all qemu targets except qemuppc, qemuarm Mar 23 20:57:09 I would like to disable some systemd services , like systemd-timesyncd, how is the best way to do this in my image's recipe? Mar 23 20:58:26 I will need create a .bbappend to remove this services? Mar 23 21:06:53 khem: we don't have the autobuilders for that. We don't do world on all arches for glibc, let alone musl Mar 23 21:07:10 khem: we'd need major AB funds to make something like that happen Mar 23 21:08:07 When creating a new layer for making my own image based off of core-image-minimal, what should I select my priority to be? Mar 23 21:08:45 Should it match that of core-image-minimals specified priority in poky/meta/conflayer.conf? Mar 23 21:09:38 core-image-minimal doesn't have a priority, the layer does. and it doesn't matter if the layer is just providing a new recipe with a new name. layer priority controls overrides when the same file is in both layers, or recipes with the same name, and controls bbappend application order Mar 23 21:13:43 kergoth: Instead of copying core-image-minimal.bb into my new layer and renaming it riz-image.bb, why can't I just create riz-image.bb in my new layer and put 'images/requires core-image-minimal.bb' Mar 23 21:14:05 'requires images/core-image-minimal.bb', sorry Mar 23 21:14:10 you could do that, yes, though that particular syntax is wrong Mar 23 21:15:04 do you really want to inherit any new packages upstream adds to the image, when your purpose is entirely different from that image? Mar 23 21:15:09 doesn't seem wise to me Mar 23 21:15:33 Fair enough Mar 23 21:15:34 core-image-* are *examples* Mar 23 21:15:39 besides there's really not that much in core-image-minimal that you wouldn't want to change Mar 23 21:15:52 Then that brings me back to the priority question Mar 23 21:15:59 I guess I did not fully understand Mar 23 21:16:08 as kergoth said it doesn't matter in this situation Mar 23 21:16:13 If it isnt inheriting a priority, what is overriding it? Mar 23 21:16:19 Ok Mar 23 21:16:37 that question doesn't make sense Mar 23 21:16:43 'inheriting a priority'? what does that even mean? Mar 23 21:16:46 what suggests to you that it's being overridden? Mar 23 21:16:47 'overriding it'? Mar 23 21:17:00 I guess I misunderstand Mar 23 21:17:09 I am used to inheriting priorities in WEC7 Mar 23 21:17:22 Again, I am ignorant to yocto/linux Mar 23 21:17:32 no idea what that even is. re-read my response, i just told you what layer priorities are used for Mar 23 21:17:54 to generally answer your question the priority would be set highest for your custom layer since that's where you would want things to take precedence Mar 23 21:18:12 That was what I was looking for Mar 23 21:18:13 Thanks Mar 23 21:18:14 something like 10 perhaps (but the exact number isn't really important) Mar 23 21:18:27 I see Mar 23 21:19:04 And would I have to copy any files taken from poky/meta/recipes-core in my new layer? Mar 23 21:19:17 only the image recipe Mar 23 21:19:44 How does it know to pull the packages from elsewhere? I am missing where that link is Mar 23 21:20:08 I mean as far as knowing where to look for the packages in the file system Mar 23 21:20:33 or recipes, not packages. Sorrry Mar 23 21:25:02 riz__: they're all collectively pointed to by your bblayers.conf Mar 23 21:25:17 Ahh. Perfect Mar 23 21:25:44 or perhaps more strictly, bblayers.conf points to each layer directory, in each of which a conf/layer.conf exists that specifies how to find the recipes Mar 23 21:27:06 Understood. Thx Mar 23 21:34:05 damnit, why is devshell still not working for me Mar 23 21:34:33 because it hates you Mar 23 21:34:36 just prints its attempts to run screen -r repeatedly without actually connecting Mar 23 21:34:59 hasn't worked for me on these machines in at least 6 months, but i'd thought it was the use of the chroot. not using it anymore.. Mar 23 21:36:21 is screen devshell working for others right now? Mar 23 22:00:00 kergoth: seems to be working here Mar 23 22:06:33 hmmm, strange Mar 23 22:06:43 this is stock ubuntu 14.04 Mar 23 22:28:36 armpit, thanks for nagging about signing up for OEDAM Mar 23 22:29:15 np. with only 10 people we can get a party bus Mar 23 22:31:24 limo rides **** ENDING LOGGING AT Thu Mar 24 02:59:58 2016